Skip to contents

Classify the measurement scale of an outcome variable

Source: R/mixed-models.R
classify_outcome.Rd

Decides how a dependent variable should be modelled by inspecting its type and distribution of values. The measurement scale is the first branch of a principled model choice: it dictates the family (Gaussian, binomial, Poisson, cumulative-link) before any distributional assumption is checked.

Usage

classify_outcome(y, ordinal_max_levels = 7L)

Arguments

y

The outcome vector.

ordinal_max_levels

Integer. Integer-valued numerics with at most this many distinct values are treated as ordinal (Likert-like). Default 7.

Value

A single string, one of "continuous", "ordinal", "binary", "count", or "nominal".

Details

The rules are deliberately simple and transparent:

  • ordered factor \(\rightarrow\) "ordinal";

  • logical, a two-level factor, or a numeric/character with exactly two distinct values \(\rightarrow\) "binary";

  • unordered factor/character with more than two levels \(\rightarrow\) "nominal";

  • integer-valued numeric with at most ordinal_max_levels distinct values (a Likert-type item) \(\rightarrow\) "ordinal";

  • non-negative integer-valued numeric with more distinct values \(\rightarrow\) "count";

  • any other numeric \(\rightarrow\) "continuous".

The heuristics can never be perfect (a 1–7 Likert item and a small count are genuinely ambiguous); pass an explicit outcome_type to recommend_test() when you want to override them.

Examples

classify_outcome(rnorm(50)) # "continuous"
#> [1] "continuous"
classify_outcome(factor(sample(1:5, 50, TRUE), ordered = TRUE)) # "ordinal"
#> [1] "ordinal"
classify_outcome(sample(0:1, 50, TRUE)) # "binary"
#> [1] "binary"
classify_outcome(rpois(50, 3)) # "count"
#> [1] "count"